def train_dialogue_model(domain_file,
stories_file,
output_path,
interpreter=None,
endpoints=AvailableEndpoints(),
max_history=None,
dump_flattened_stories=False,
policy_config=None,
kwargs=None):
if not kwargs:
kwargs = {}
fallback_args, kwargs = utils.extract_args(
kwargs, {"nlu_threshold", "core_threshold", "fallback_action_name"})
policies = config.load(policy_config, fallback_args, max_history)
agent = Agent(domain_file,
generator=endpoints.nlg,
action_endpoint=endpoints.action,
interpreter=interpreter,
policies=policies)
data_load_args, kwargs = utils.extract_args(
kwargs, {
"use_story_concatenation", "unique_last_num_states",
"augmentation_factor", "remove_duplicates", "debug_plots"
})
training_data = agent.load_data(stories_file, **data_load_args)
agent.train(training_data, **kwargs)
agent.persist(output_path, dump_flattened_stories)
return agent
def train_dialogue_model(domain_file,
stories_file,
output_path,
interpreter=None,
endpoints=AvailableEndpoints(),
dump_stories=False,
policy_config=None,
exclusion_percentage=None,
kwargs=None):
if not kwargs:
kwargs = {}
policies = config.load(policy_config)
agent = Agent(domain_file,
generator=endpoints.nlg,
action_endpoint=endpoints.action,
interpreter=interpreter,
policies=policies)
data_load_args, kwargs = utils.extract_args(
kwargs, {
"use_story_concatenation", "unique_last_num_states",
"augmentation_factor", "remove_duplicates", "debug_plots"
})
training_data = agent.load_data(stories_file,
exclusion_percentage=exclusion_percentage,
**data_load_args)
agent.train(training_data, **kwargs)
agent.persist(output_path, dump_stories)
return agent
def train(domain_file: Text,
stories_file: Text,
output_path: Text,
interpreter: Optional[NaturalLanguageInterpreter] = None,
endpoints: AvailableEndpoints = AvailableEndpoints(),
dump_stories: bool = False,
policy_config: Text = None,
exclusion_percentage: int = None,
kwargs: Optional[Dict] = None):
from rasa_core.agent import Agent
if not kwargs:
kwargs = {}
policies = config.load(policy_config)
agent = Agent(domain_file,
generator=endpoints.nlg,
action_endpoint=endpoints.action,
interpreter=interpreter,
policies=policies)
data_load_args, kwargs = utils.extract_args(
kwargs, {
"use_story_concatenation", "unique_last_num_states",
"augmentation_factor", "remove_duplicates", "debug_plots"
})
training_data = agent.load_data(stories_file,
exclusion_percentage=exclusion_percentage,
**data_load_args)
agent.train(training_data, **kwargs)
agent.persist(output_path, dump_stories)
return agent
def train_dialogue_model(domain_file, stories_file, output_path,
interpreter=None,
endpoints=AvailableEndpoints(),
max_history=None,
dump_flattened_stories=False,
kwargs=None):
if not kwargs:
kwargs = {}
fallback_args, kwargs = utils.extract_args(kwargs,
{"nlu_threshold",
"core_threshold",
"fallback_action_name"})
policies = [
FallbackPolicy(
fallback_args.get("nlu_threshold",
DEFAULT_NLU_FALLBACK_THRESHOLD),
fallback_args.get("core_threshold",
DEFAULT_CORE_FALLBACK_THRESHOLD),
fallback_args.get("fallback_action_name",
DEFAULT_FALLBACK_ACTION)),
MemoizationPolicy(
max_history=max_history),
KerasPolicy(
MaxHistoryTrackerFeaturizer(BinarySingleStateFeaturizer(),
max_history=max_history))]
agent = Agent(domain_file,
generator=endpoints.nlg,
action_endpoint=endpoints.action,
interpreter=interpreter,
policies=policies)
data_load_args, kwargs = utils.extract_args(kwargs,
{"use_story_concatenation",
"unique_last_num_states",
"augmentation_factor",
"remove_duplicates",
"debug_plots"})
training_data = agent.load_data(stories_file, **data_load_args)
agent.train(training_data, **kwargs)
agent.persist(output_path, dump_flattened_stories)
return agent
def do_interactive_learning(cmdline_args, stories, additional_arguments):
_endpoints = AvailableEndpoints.read_endpoints(cmdline_args.endpoints)
_interpreter = NaturalLanguageInterpreter.create(cmdline_args.nlu,
_endpoints.nlu)
if cmdline_args.core:
if cmdline_args.finetune:
raise ValueError("--core can only be used without "
"--finetune flag.")
logger.info("Loading a pre-trained model. This means that "
"all training-related parameters will be ignored.")
_broker = PikaProducer.from_endpoint_config(_endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(
None, _endpoints.tracker_store, _broker)
_agent = Agent.load(cmdline_args.core,
interpreter=_interpreter,
generator=_endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=_endpoints.action)
else:
if cmdline_args.out:
model_directory = cmdline_args.out
else:
model_directory = tempfile.mkdtemp(suffix="_core_model")
_agent = train_dialogue_model(cmdline_args.domain, stories,
model_directory, _interpreter,
_endpoints, cmdline_args.dump_stories,
cmdline_args.config[0], None,
additional_arguments)
interactive.run_interactive_learning(
_agent,
stories,
finetune=cmdline_args.finetune,
skip_visualization=cmdline_args.skip_visualization)
def do_interactive_learning(cmdline_args, stories, additional_arguments):
_endpoints = AvailableEndpoints.read_endpoints(cmdline_args.endpoints)
_interpreter = NaturalLanguageInterpreter.create(cmdline_args.nlu,
_endpoints.nlu)
if (isinstance(cmdline_args.config, list)
and len(cmdline_args.config) > 1):
raise ValueError("You can only pass one config file at a time")
if cmdline_args.core and cmdline_args.finetune:
raise ValueError("--core can only be used without --finetune flag.")
elif cmdline_args.core:
logger.info("loading a pre-trained model. "
"all training-related parameters will be ignored")
_broker = PikaProducer.from_endpoint_config(_endpoints.event_broker)
_tracker_store = TrackerStore.find_tracker_store(
None, _endpoints.tracker_store, _broker)
_agent = Agent.load(cmdline_args.core,
interpreter=_interpreter,
generator=_endpoints.nlg,
tracker_store=_tracker_store,
action_endpoint=_endpoints.action)
else:
if not cmdline_args.out:
raise ValueError("you must provide a path where the model "
"will be saved using -o / --out")
_agent = train_dialogue_model(cmdline_args.domain, stories,
cmdline_args.out, _interpreter,
_endpoints, cmdline_args.dump_stories,
cmdline_args.config[0], None,
additional_arguments)
interactive.run_interactive_learning(
_agent,
stories,
finetune=cmdline_args.finetune,
skip_visualization=cmdline_args.skip_visualization)
"epochs": cmdline_args.epochs,
"batch_size": cmdline_args.batch_size,
"validation_split": cmdline_args.validation_split,
"augmentation_factor": cmdline_args.augmentation,
"debug_plots": cmdline_args.debug_plots,
"nlu_threshold": cmdline_args.nlu_threshold,
"core_threshold": cmdline_args.core_threshold,
"fallback_action_name": cmdline_args.fallback_action_name
}
if cmdline_args.url:
stories = utils.download_file_from_url(cmdline_args.url)
else:
stories = cmdline_args.stories
_endpoints = AvailableEndpoints.read_endpoints(cmdline_args.endpoints)
_interpreter = NaturalLanguageInterpreter.create(cmdline_args.nlu,
_endpoints.nlu)
if cmdline_args.core:
if not cmdline_args.interactive:
raise ValueError("--core can only be used together with the"
"--interactive flag.")
elif cmdline_args.finetune:
raise ValueError("--core can only be used together with the"
"--interactive flag and without --finetune flag.")
else:
logger.info(
"loading a pre-trained model. all training-related parameters will be ignored"
)
_agent = Agent.load(cmdline_args.core,
